Split Bregman Method, Image Decomposition

Image decomposition is a fundamental problem in image processing, which decomposed the different components of the image by decomposition method to extract the required image information. Early image decomposition processing technology is mainly based on the variational methods of partial differential equations and the dual method to decompose the image, in recent years, based on the Split Bregman method of image processing technology to become a research hotspot. In this paper, we combination of variational method and Split Bregman method to do a more in-depth study for image decomposition technique, the main work in the following areas:First, analysis of the solution process for the Split Bregman method by detailed, and master the application in the TV mode, and then extended to other models. Second, research the application of image decomposition based on Split Bregman TV models and TV-L1 model, and verified by experiments which not only has better denoising effect, but also a better results to decompose the structure part and the texture part of the image. Third, focus on the image decomposition model of VO model and OSV model by introducted of Split Bregman method, also proposed an improved decomposition model of L1-H-1 model based on TV-L1 model and OSV model, and then, combined with Split Bregman method to achieved. Finally, through the experiments to verified the Split Bregman method have the rapid convergence efficiency and the better effects of decomposition, while also verified the improved method can be very good for image decomposition and image denoising, and compared with the traditional method has a better computing efficiency. Fourth, introduce the application of the image decomposition in image processing, analysis and verified its application in the texture image segmentation and image restoration by experimental.